// SPDX-License-Identifier: GPL-2.0
#include "bcachefs.h"
+#include "backpointers.h"
#include "bkey_buf.h"
#include "alloc_background.h"
#include "btree_gc.h"
#include "buckets.h"
#include "dirent.h"
#include "ec.h"
+#include "errcode.h"
#include "error.h"
#include "fs-common.h"
#include "fsck.h"
return keys->d + idx_to_pos(keys, idx);
}
-size_t bch2_journal_key_search(struct journal_keys *keys,
- enum btree_id id, unsigned level,
- struct bpos pos)
+static size_t __bch2_journal_key_search(struct journal_keys *keys,
+ enum btree_id id, unsigned level,
+ struct bpos pos)
{
size_t l = 0, r = keys->nr, m;
BUG_ON(l &&
__journal_key_cmp(id, level, pos, idx_to_key(keys, l - 1)) <= 0);
- return idx_to_pos(keys, l);
+ return l;
}
-struct bkey_i *bch2_journal_keys_peek(struct bch_fs *c, enum btree_id btree_id,
- unsigned level, struct bpos pos)
+static size_t bch2_journal_key_search(struct journal_keys *keys,
+ enum btree_id id, unsigned level,
+ struct bpos pos)
+{
+ return idx_to_pos(keys, __bch2_journal_key_search(keys, id, level, pos));
+}
+
+struct bkey_i *bch2_journal_keys_peek_upto(struct bch_fs *c, enum btree_id btree_id,
+ unsigned level, struct bpos pos,
+ struct bpos end_pos, size_t *idx)
{
struct journal_keys *keys = &c->journal_keys;
- size_t idx = bch2_journal_key_search(keys, btree_id, level, pos);
+ unsigned iters = 0;
+ struct journal_key *k;
+search:
+ if (!*idx)
+ *idx = __bch2_journal_key_search(keys, btree_id, level, pos);
+
+ while ((k = *idx < keys->nr ? idx_to_key(keys, *idx) : NULL)) {
+ if (__journal_key_cmp(btree_id, level, end_pos, k) < 0)
+ return NULL;
+
+ if (__journal_key_cmp(btree_id, level, pos, k) <= 0 &&
+ !k->overwritten)
+ return k->k;
- while (idx < keys->size &&
- keys->d[idx].overwritten) {
- idx++;
- if (idx == keys->gap)
- idx += keys->size - keys->nr;
+ (*idx)++;
+ iters++;
+ if (iters == 10) {
+ *idx = 0;
+ goto search;
+ }
}
- if (idx < keys->size &&
- keys->d[idx].btree_id == btree_id &&
- keys->d[idx].level == level)
- return keys->d[idx].k;
return NULL;
}
+struct bkey_i *bch2_journal_keys_peek_slot(struct bch_fs *c, enum btree_id btree_id,
+ unsigned level, struct bpos pos)
+{
+ size_t idx = 0;
+
+ return bch2_journal_keys_peek_upto(c, btree_id, level, pos, pos, &idx);
+}
+
static void journal_iters_fix(struct bch_fs *c)
{
struct journal_keys *keys = &c->journal_keys;
/* The key we just inserted is immediately before the gap: */
- struct journal_key *n = &keys->d[keys->gap - 1];
size_t gap_end = keys->gap + (keys->size - keys->nr);
struct btree_and_journal_iter *iter;
/*
- * If an iterator points one after the key we just inserted,
- * and the key we just inserted compares >= the iterator's position,
- * decrement the iterator so it points at the key we just inserted:
+ * If an iterator points one after the key we just inserted, decrement
+ * the iterator so it points at the key we just inserted - if the
+ * decrement was unnecessary, bch2_btree_and_journal_iter_peek() will
+ * handle that:
*/
list_for_each_entry(iter, &c->journal_iters, journal.list)
- if (iter->journal.idx == gap_end &&
- iter->last &&
- iter->b->c.btree_id == n->btree_id &&
- iter->b->c.level == n->level &&
- bpos_cmp(n->k->k.p, iter->unpacked.p) >= 0)
+ if (iter->journal.idx == gap_end)
iter->journal.idx = keys->gap - 1;
}
struct journal_keys new_keys = {
.nr = keys->nr,
.size = max_t(size_t, keys->size, 8) * 2,
- .journal_seq_base = keys->journal_seq_base,
};
- new_keys.d = kvmalloc(sizeof(new_keys.d[0]) * new_keys.size, GFP_KERNEL);
+ new_keys.d = kvmalloc_array(new_keys.size, sizeof(new_keys.d[0]), GFP_KERNEL);
if (!new_keys.d) {
bch_err(c, "%s: error allocating new key array (size %zu)",
__func__, new_keys.size);
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_journal_key_insert;
}
/* Since @keys was full, there was no gap: */
n = kmalloc(bkey_bytes(&k->k), GFP_KERNEL);
if (!n)
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_journal_key_insert;
bkey_copy(n, k);
ret = bch2_journal_key_insert_take(c, id, level, n);
if (idx < keys->size &&
keys->d[idx].btree_id == btree &&
keys->d[idx].level == level &&
- !bpos_cmp(keys->d[idx].k->k.p, pos))
+ bpos_eq(keys->d[idx].k->k.p, pos))
keys->d[idx].overwritten = true;
}
-static struct bkey_i *bch2_journal_iter_peek(struct journal_iter *iter)
+static void bch2_journal_iter_advance(struct journal_iter *iter)
+{
+ if (iter->idx < iter->keys->size) {
+ iter->idx++;
+ if (iter->idx == iter->keys->gap)
+ iter->idx += iter->keys->size - iter->keys->nr;
+ }
+}
+
+struct bkey_s_c bch2_journal_iter_peek(struct journal_iter *iter)
{
struct journal_key *k = iter->keys->d + iter->idx;
- while (k < iter->keys->d + iter->keys->nr &&
+ while (k < iter->keys->d + iter->keys->size &&
k->btree_id == iter->btree_id &&
k->level == iter->level) {
if (!k->overwritten)
- return k->k;
+ return bkey_i_to_s_c(k->k);
- iter->idx++;
+ bch2_journal_iter_advance(iter);
k = iter->keys->d + iter->idx;
}
- return NULL;
-}
-
-static void bch2_journal_iter_advance(struct journal_iter *iter)
-{
- if (iter->idx < iter->keys->size) {
- iter->idx++;
- if (iter->idx == iter->keys->gap)
- iter->idx += iter->keys->size - iter->keys->nr;
- }
+ return bkey_s_c_null;
}
static void bch2_journal_iter_exit(struct journal_iter *iter)
void bch2_btree_and_journal_iter_advance(struct btree_and_journal_iter *iter)
{
- switch (iter->last) {
- case none:
- break;
- case btree:
- bch2_journal_iter_advance_btree(iter);
- break;
- case journal:
- bch2_journal_iter_advance(&iter->journal);
- break;
- }
-
- iter->last = none;
+ if (bpos_eq(iter->pos, SPOS_MAX))
+ iter->at_end = true;
+ else
+ iter->pos = bpos_successor(iter->pos);
}
struct bkey_s_c bch2_btree_and_journal_iter_peek(struct btree_and_journal_iter *iter)
{
- struct bkey_s_c ret;
+ struct bkey_s_c btree_k, journal_k, ret;
+again:
+ if (iter->at_end)
+ return bkey_s_c_null;
- while (1) {
- struct bkey_s_c btree_k =
- bch2_journal_iter_peek_btree(iter);
- struct bkey_s_c journal_k =
- bkey_i_to_s_c(bch2_journal_iter_peek(&iter->journal));
-
- if (btree_k.k && journal_k.k) {
- int cmp = bpos_cmp(btree_k.k->p, journal_k.k->p);
+ while ((btree_k = bch2_journal_iter_peek_btree(iter)).k &&
+ bpos_lt(btree_k.k->p, iter->pos))
+ bch2_journal_iter_advance_btree(iter);
- if (!cmp)
- bch2_journal_iter_advance_btree(iter);
+ while ((journal_k = bch2_journal_iter_peek(&iter->journal)).k &&
+ bpos_lt(journal_k.k->p, iter->pos))
+ bch2_journal_iter_advance(&iter->journal);
- iter->last = cmp < 0 ? btree : journal;
- } else if (btree_k.k) {
- iter->last = btree;
- } else if (journal_k.k) {
- iter->last = journal;
- } else {
- iter->last = none;
- return bkey_s_c_null;
- }
+ ret = journal_k.k &&
+ (!btree_k.k || bpos_le(journal_k.k->p, btree_k.k->p))
+ ? journal_k
+ : btree_k;
- ret = iter->last == journal ? journal_k : btree_k;
+ if (ret.k && iter->b && bpos_gt(ret.k->p, iter->b->data->max_key))
+ ret = bkey_s_c_null;
- if (iter->b &&
- bpos_cmp(ret.k->p, iter->b->data->max_key) > 0) {
- iter->journal.idx = iter->journal.keys->nr;
- iter->last = none;
- return bkey_s_c_null;
+ if (ret.k) {
+ iter->pos = ret.k->p;
+ if (bkey_deleted(ret.k)) {
+ bch2_btree_and_journal_iter_advance(iter);
+ goto again;
}
-
- if (!bkey_deleted(ret.k))
- break;
-
- bch2_btree_and_journal_iter_advance(iter);
+ } else {
+ iter->pos = SPOS_MAX;
+ iter->at_end = true;
}
return ret;
}
-struct bkey_s_c bch2_btree_and_journal_iter_next(struct btree_and_journal_iter *iter)
-{
- bch2_btree_and_journal_iter_advance(iter);
-
- return bch2_btree_and_journal_iter_peek(iter);
-}
-
void bch2_btree_and_journal_iter_exit(struct btree_and_journal_iter *iter)
{
bch2_journal_iter_exit(&iter->journal);
iter->node_iter = node_iter;
bch2_journal_iter_init(c, &iter->journal, b->c.btree_id, b->c.level, pos);
INIT_LIST_HEAD(&iter->journal.list);
+ iter->pos = b->data->min_key;
+ iter->at_end = false;
}
/*
keys->nr = keys->gap = keys->size = 0;
}
+static void __journal_keys_sort(struct journal_keys *keys)
+{
+ struct journal_key *src, *dst;
+
+ sort(keys->d, keys->nr, sizeof(keys->d[0]), journal_sort_key_cmp, NULL);
+
+ src = dst = keys->d;
+ while (src < keys->d + keys->nr) {
+ while (src + 1 < keys->d + keys->nr &&
+ src[0].btree_id == src[1].btree_id &&
+ src[0].level == src[1].level &&
+ bpos_eq(src[0].k->k.p, src[1].k->k.p))
+ src++;
+
+ *dst++ = *src++;
+ }
+
+ keys->nr = dst - keys->d;
+}
+
static int journal_keys_sort(struct bch_fs *c)
{
struct genradix_iter iter;
struct journal_replay *i, **_i;
struct jset_entry *entry;
- struct bkey_i *k, *_n;
+ struct bkey_i *k;
struct journal_keys *keys = &c->journal_keys;
- struct journal_key *src, *dst;
- size_t nr_keys = 0;
+ size_t nr_keys = 0, nr_read = 0;
genradix_for_each(&c->journal_entries, iter, _i) {
i = *_i;
if (!i || i->ignore)
continue;
- if (!keys->journal_seq_base)
- keys->journal_seq_base = le64_to_cpu(i->j.seq);
-
- for_each_jset_key(k, _n, entry, &i->j)
+ for_each_jset_key(k, entry, &i->j)
nr_keys++;
}
keys->size = roundup_pow_of_two(nr_keys);
- keys->d = kvmalloc(sizeof(keys->d[0]) * keys->size, GFP_KERNEL);
- if (!keys->d)
- return -ENOMEM;
+ keys->d = kvmalloc_array(keys->size, sizeof(keys->d[0]), GFP_KERNEL);
+ if (!keys->d) {
+ bch_err(c, "Failed to allocate buffer for sorted journal keys (%zu keys); trying slowpath",
+ nr_keys);
+
+ do {
+ keys->size >>= 1;
+ keys->d = kvmalloc_array(keys->size, sizeof(keys->d[0]), GFP_KERNEL);
+ } while (!keys->d && keys->size > nr_keys / 8);
+
+ if (!keys->d) {
+ bch_err(c, "Failed to allocate %zu size buffer for sorted journal keys; exiting",
+ keys->size);
+ return -BCH_ERR_ENOMEM_journal_keys_sort;
+ }
+ }
genradix_for_each(&c->journal_entries, iter, _i) {
i = *_i;
if (!i || i->ignore)
continue;
- BUG_ON(le64_to_cpu(i->j.seq) - keys->journal_seq_base > U32_MAX);
+ cond_resched();
+
+ for_each_jset_key(k, entry, &i->j) {
+ if (keys->nr == keys->size) {
+ __journal_keys_sort(keys);
+
+ if (keys->nr > keys->size * 7 / 8) {
+ bch_err(c, "Too many journal keys for slowpath; have %zu compacted, buf size %zu, processed %zu/%zu",
+ keys->nr, keys->size, nr_read, nr_keys);
+ return -BCH_ERR_ENOMEM_journal_keys_sort;
+ }
+ }
- for_each_jset_key(k, _n, entry, &i->j)
keys->d[keys->nr++] = (struct journal_key) {
.btree_id = entry->btree_id,
.level = entry->level,
.k = k,
- .journal_seq = le64_to_cpu(i->j.seq) -
- keys->journal_seq_base,
+ .journal_seq = le64_to_cpu(i->j.seq),
.journal_offset = k->_data - i->j._data,
};
- }
-
- sort(keys->d, keys->nr, sizeof(keys->d[0]), journal_sort_key_cmp, NULL);
-
- src = dst = keys->d;
- while (src < keys->d + keys->nr) {
- while (src + 1 < keys->d + keys->nr &&
- src[0].btree_id == src[1].btree_id &&
- src[0].level == src[1].level &&
- !bpos_cmp(src[0].k->k.p, src[1].k->k.p))
- src++;
- *dst++ = *src++;
+ nr_read++;
+ }
}
- keys->nr = dst - keys->d;
+ __journal_keys_sort(keys);
keys->gap = keys->nr;
+
+ bch_verbose(c, "Journal keys: %zu read, %zu after sorting and compacting", nr_keys, keys->nr);
return 0;
}
static void replay_now_at(struct journal *j, u64 seq)
{
BUG_ON(seq < j->replay_journal_seq);
- BUG_ON(seq > j->replay_journal_seq_end);
+
+ seq = min(seq, j->replay_journal_seq_end);
while (j->replay_journal_seq < seq)
bch2_journal_pin_put(j, j->replay_journal_seq++);
return cmp_int(l->journal_seq, r->journal_seq);
}
-static int bch2_journal_replay(struct bch_fs *c)
+static int bch2_journal_replay(struct bch_fs *c, u64 start_seq, u64 end_seq)
{
struct journal_keys *keys = &c->journal_keys;
struct journal_key **keys_sorted, *k;
keys_sorted = kvmalloc_array(sizeof(*keys_sorted), keys->nr, GFP_KERNEL);
if (!keys_sorted)
- return -ENOMEM;
+ return -BCH_ERR_ENOMEM_journal_replay;
for (i = 0; i < keys->nr; i++)
keys_sorted[i] = &keys->d[i];
sizeof(keys_sorted[0]),
journal_sort_seq_cmp, NULL);
- if (keys->nr)
- replay_now_at(j, keys->journal_seq_base);
+ if (keys->nr) {
+ ret = bch2_journal_log_msg(c, "Starting journal replay (%zu keys in entries %llu-%llu)",
+ keys->nr, start_seq, end_seq);
+ if (ret)
+ goto err;
+ }
for (i = 0; i < keys->nr; i++) {
k = keys_sorted[i];
cond_resched();
- if (!k->allocated)
- replay_now_at(j, keys->journal_seq_base + k->journal_seq);
+ replay_now_at(j, k->journal_seq);
ret = bch2_trans_do(c, NULL, NULL,
BTREE_INSERT_LAZY_RW|
: 0),
bch2_journal_replay_key(&trans, k));
if (ret) {
- bch_err(c, "journal replay: error %d while replaying key at btree %s level %u",
- ret, bch2_btree_ids[k->btree_id], k->level);
+ bch_err(c, "journal replay: error while replaying key at btree %s level %u: %s",
+ bch2_btree_ids[k->btree_id], k->level, bch2_err_str(ret));
goto err;
}
}
ret = bch2_journal_error(j);
if (keys->nr && !ret)
- bch2_journal_log_msg(&c->journal, "journal replay finished");
+ bch2_journal_log_msg(c, "journal replay finished");
err:
kvfree(keys_sorted);
+
+ if (ret)
+ bch_err_fn(c, ret);
return ret;
}
if (k1)
bch2_bkey_val_to_text(&buf1, c, bkey_i_to_s_c(k1));
else
- pr_buf(&buf1, "(none)");
+ prt_printf(&buf1, "(none)");
if (k2)
bch2_bkey_val_to_text(&buf2, c, bkey_i_to_s_c(k2));
else
- pr_buf(&buf2, "(none)");
+ prt_printf(&buf2, "(none)");
mustfix_fsck_err_on(!k1 || !k2 ||
IS_ERR(k1) ||
IS_ERR(k2) ||
k1->k.u64s != k2->k.u64s ||
- memcmp(k1, k2, bkey_bytes(k1)) ||
+ memcmp(k1, k2, bkey_bytes(&k1->k)) ||
l1 != l2, c,
"superblock btree root %u doesn't match journal after clean shutdown\n"
"sb: l=%u %s\n"
GFP_KERNEL);
if (!clean) {
mutex_unlock(&c->sb_lock);
- return ERR_PTR(-ENOMEM);
+ return ERR_PTR(-BCH_ERR_ENOMEM_read_superblock_clean);
}
ret = bch2_sb_clean_validate_late(c, clean, READ);
return ERR_PTR(ret);
}
+static bool btree_id_is_alloc(enum btree_id id)
+{
+ switch (id) {
+ case BTREE_ID_alloc:
+ case BTREE_ID_backpointers:
+ case BTREE_ID_need_discard:
+ case BTREE_ID_freespace:
+ case BTREE_ID_bucket_gens:
+ return true;
+ default:
+ return false;
+ }
+}
+
static int read_btree_roots(struct bch_fs *c)
{
unsigned i;
if (!r->alive)
continue;
- if (i == BTREE_ID_alloc &&
+ if (btree_id_is_alloc(i) &&
c->opts.reconstruct_alloc) {
c->sb.compat &= ~(1ULL << BCH_COMPAT_alloc_info);
continue;
}
if (r->error) {
- __fsck_err(c, i == BTREE_ID_alloc
+ __fsck_err(c, btree_id_is_alloc(i)
? FSCK_CAN_IGNORE : 0,
"invalid btree root %s",
bch2_btree_ids[i]);
ret = bch2_btree_root_read(c, i, &r->key, r->level);
if (ret) {
- __fsck_err(c, i == BTREE_ID_alloc
+ __fsck_err(c,
+ btree_id_is_alloc(i)
? FSCK_CAN_IGNORE : 0,
"error reading btree root %s",
bch2_btree_ids[i]);
- if (i == BTREE_ID_alloc)
+ if (btree_id_is_alloc(i))
c->sb.compat &= ~(1ULL << BCH_COMPAT_alloc_info);
}
}
- for (i = 0; i < BTREE_ID_NR; i++)
- if (!c->btree_roots[i].b)
+ for (i = 0; i < BTREE_ID_NR; i++) {
+ struct btree_root *r = &c->btree_roots[i];
+
+ if (!r->b) {
+ r->alive = false;
+ r->level = 0;
bch2_btree_root_alloc(c, i);
+ }
+ }
fsck_err:
return ret;
}
static int bch2_fs_initialize_subvolumes(struct bch_fs *c)
{
- struct bkey_i_snapshot root_snapshot;
- struct bkey_i_subvolume root_volume;
+ struct bkey_i_snapshot_tree root_tree;
+ struct bkey_i_snapshot root_snapshot;
+ struct bkey_i_subvolume root_volume;
int ret;
+ bkey_snapshot_tree_init(&root_tree.k_i);
+ root_tree.k.p.offset = 1;
+ root_tree.v.master_subvol = cpu_to_le32(1);
+ root_tree.v.root_snapshot = cpu_to_le32(U32_MAX);
+
bkey_snapshot_init(&root_snapshot.k_i);
root_snapshot.k.p.offset = U32_MAX;
root_snapshot.v.flags = 0;
root_snapshot.v.parent = 0;
root_snapshot.v.subvol = BCACHEFS_ROOT_SUBVOL;
- root_snapshot.v.pad = 0;
+ root_snapshot.v.tree = cpu_to_le32(1);
SET_BCH_SNAPSHOT_SUBVOL(&root_snapshot.v, true);
- ret = bch2_btree_insert(c, BTREE_ID_snapshots,
- &root_snapshot.k_i,
- NULL, NULL, 0);
- if (ret)
- return ret;
-
-
bkey_subvolume_init(&root_volume.k_i);
root_volume.k.p.offset = BCACHEFS_ROOT_SUBVOL;
root_volume.v.flags = 0;
root_volume.v.snapshot = cpu_to_le32(U32_MAX);
root_volume.v.inode = cpu_to_le64(BCACHEFS_ROOT_INO);
- ret = bch2_btree_insert(c, BTREE_ID_subvolumes,
- &root_volume.k_i,
- NULL, NULL, 0);
+ ret = bch2_btree_insert(c, BTREE_ID_snapshot_trees,
+ &root_tree.k_i,
+ NULL, NULL, 0) ?:
+ bch2_btree_insert(c, BTREE_ID_snapshots,
+ &root_snapshot.k_i,
+ NULL, NULL, 0) ?:
+ bch2_btree_insert(c, BTREE_ID_subvolumes,
+ &root_volume.k_i,
+ NULL, NULL, 0);
if (ret)
- return ret;
-
- return 0;
+ bch_err_fn(c, ret);
+ return ret;
}
-static int bch2_fs_upgrade_for_subvolumes(struct btree_trans *trans)
+static int __bch2_fs_upgrade_for_subvolumes(struct btree_trans *trans)
{
struct btree_iter iter;
struct bkey_s_c k;
struct bch_inode_unpacked inode;
int ret;
- bch2_trans_iter_init(trans, &iter, BTREE_ID_inodes,
- SPOS(0, BCACHEFS_ROOT_INO, U32_MAX), 0);
- k = bch2_btree_iter_peek_slot(&iter);
+ k = bch2_bkey_get_iter(trans, &iter, BTREE_ID_inodes,
+ SPOS(0, BCACHEFS_ROOT_INO, U32_MAX), 0);
ret = bkey_err(k);
if (ret)
- goto err;
+ return ret;
if (!bkey_is_inode(k.k)) {
bch_err(trans->c, "root inode not found");
- ret = -ENOENT;
+ ret = -BCH_ERR_ENOENT_inode;
goto err;
}
return ret;
}
+/* set bi_subvol on root inode */
+noinline_for_stack
+static int bch2_fs_upgrade_for_subvolumes(struct bch_fs *c)
+{
+ int ret = bch2_trans_do(c, NULL, NULL, BTREE_INSERT_LAZY_RW,
+ __bch2_fs_upgrade_for_subvolumes(&trans));
+ if (ret)
+ bch_err_fn(c, ret);
+ return ret;
+}
+
int bch2_fs_recovery(struct bch_fs *c)
{
- const char *err = "cannot allocate memory";
struct bch_sb_field_clean *clean = NULL;
struct jset *last_journal_entry = NULL;
- u64 blacklist_seq, journal_seq;
+ u64 last_seq, blacklist_seq, journal_seq;
bool write_sb = false;
int ret = 0;
goto err;
}
- if (!(c->sb.features & (1ULL << BCH_FEATURE_alloc_v2))) {
- bch_info(c, "alloc_v2 feature bit not set, fsck required");
- c->opts.fsck = true;
- c->opts.fix_errors = FSCK_OPT_YES;
- }
-
- if (!c->replicas.entries ||
- c->opts.rebuild_replicas) {
- bch_info(c, "building replicas info");
- set_bit(BCH_FS_REBUILD_REPLICAS, &c->flags);
- }
-
if (!c->opts.nochanges) {
- if (c->sb.version < bcachefs_metadata_version_new_data_types) {
- bch_info(c, "version prior to new_data_types, upgrade and fsck required");
+ if (c->sb.version < bcachefs_metadata_required_upgrade_below) {
+ bch_info(c, "version %s (%u) prior to %s (%u), upgrade and fsck required",
+ bch2_metadata_versions[c->sb.version],
+ c->sb.version,
+ bch2_metadata_versions[bcachefs_metadata_required_upgrade_below],
+ bcachefs_metadata_required_upgrade_below);
c->opts.version_upgrade = true;
c->opts.fsck = true;
c->opts.fix_errors = FSCK_OPT_YES;
}
}
+ if (c->opts.fsck && c->opts.norecovery) {
+ bch_err(c, "cannot select both norecovery and fsck");
+ ret = -EINVAL;
+ goto err;
+ }
+
ret = bch2_blacklist_table_initialize(c);
if (ret) {
bch_err(c, "error initializing blacklist table");
struct journal_replay **i;
bch_verbose(c, "starting journal read");
- ret = bch2_journal_read(c, &blacklist_seq, &journal_seq);
+ ret = bch2_journal_read(c, &last_seq, &blacklist_seq, &journal_seq);
if (ret)
goto err;
+ /*
+ * note: cmd_list_journal needs the blacklist table fully up to date so
+ * it can asterisk ignored journal entries:
+ */
+ if (c->opts.read_journal_only)
+ goto out;
+
genradix_for_each_reverse(&c->journal_entries, iter, i)
if (*i && !(*i)->ignore) {
last_journal_entry = &(*i)->j;
if (!last_journal_entry) {
fsck_err_on(!c->sb.clean, c, "no journal entries found");
- goto use_clean;
+ if (clean)
+ goto use_clean;
+
+ genradix_for_each_reverse(&c->journal_entries, iter, i)
+ if (*i) {
+ last_journal_entry = &(*i)->j;
+ (*i)->ignore = false;
+ break;
+ }
}
ret = journal_keys_sort(c);
use_clean:
if (!clean) {
bch_err(c, "no superblock clean section found");
- ret = BCH_FSCK_REPAIR_IMPOSSIBLE;
+ ret = -BCH_ERR_fsck_repair_impossible;
goto err;
}
blacklist_seq = journal_seq = le64_to_cpu(clean->journal_seq) + 1;
}
- if (c->opts.read_journal_only)
- goto out;
-
if (c->opts.reconstruct_alloc) {
c->sb.compat &= ~(1ULL << BCH_COMPAT_alloc_info);
drop_alloc_keys(&c->journal_keys);
journal_seq += 8;
if (blacklist_seq != journal_seq) {
- ret = bch2_journal_seq_blacklist_add(c,
+ ret = bch2_journal_log_msg(c, "blacklisting entries %llu-%llu",
+ blacklist_seq, journal_seq) ?:
+ bch2_journal_seq_blacklist_add(c,
blacklist_seq, journal_seq);
if (ret) {
bch_err(c, "error creating new journal seq blacklist entry");
}
}
- ret = bch2_fs_journal_start(&c->journal, journal_seq);
+ ret = bch2_journal_log_msg(c, "starting journal at entry %llu, replaying %llu-%llu",
+ journal_seq, last_seq, blacklist_seq - 1) ?:
+ bch2_fs_journal_start(&c->journal, journal_seq);
if (ret)
goto err;
+ if (c->opts.reconstruct_alloc)
+ bch2_journal_log_msg(c, "dropping alloc info");
+
+ /*
+ * Skip past versions that might have possibly been used (as nonces),
+ * but hadn't had their pointers written:
+ */
+ if (c->sb.encryption_type && !c->sb.clean)
+ atomic64_add(1 << 16, &c->key_version);
+
ret = read_btree_roots(c);
if (ret)
goto err;
bch_verbose(c, "starting alloc read");
- err = "error reading allocation information";
-
down_read(&c->gc_lock);
- ret = bch2_alloc_read(c);
+ ret = c->sb.version < bcachefs_metadata_version_bucket_gens
+ ? bch2_alloc_read(c)
+ : bch2_bucket_gens_read(c);
up_read(&c->gc_lock);
-
if (ret)
goto err;
bch_verbose(c, "alloc read done");
bch_verbose(c, "starting stripes_read");
- err = "error reading stripes";
ret = bch2_stripes_read(c);
if (ret)
goto err;
bch_verbose(c, "stripes_read done");
- /*
- * If we're not running fsck, this ensures bch2_fsck_err() calls are
- * instead interpreted as bch2_inconsistent_err() calls:
- */
- if (!c->opts.fsck)
- set_bit(BCH_FS_FSCK_DONE, &c->flags);
+ if (c->sb.version < bcachefs_metadata_version_snapshot_2) {
+ ret = bch2_fs_initialize_subvolumes(c);
+ if (ret)
+ goto err;
+ }
+
+ bch_verbose(c, "reading snapshots table");
+ ret = bch2_fs_snapshots_start(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "reading snapshots done");
- if (c->opts.fsck ||
- !(c->sb.compat & (1ULL << BCH_COMPAT_alloc_info)) ||
- !(c->sb.compat & (1ULL << BCH_COMPAT_alloc_metadata)) ||
- test_bit(BCH_FS_REBUILD_REPLICAS, &c->flags)) {
+ if (c->opts.fsck) {
bool metadata_only = c->opts.norecovery;
bch_info(c, "checking allocations");
- err = "error checking allocations";
ret = bch2_gc(c, true, metadata_only);
if (ret)
goto err;
bch_verbose(c, "done checking allocations");
- }
- if (c->opts.fsck) {
- bch_info(c, "checking need_discard and freespace btrees");
- err = "error checking need_discard and freespace btrees";
- ret = bch2_check_alloc_info(c);
+ set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
+
+ set_bit(BCH_FS_MAY_GO_RW, &c->flags);
+
+ bch_info(c, "starting journal replay, %zu keys", c->journal_keys.nr);
+ ret = bch2_journal_replay(c, last_seq, blacklist_seq - 1);
if (ret)
goto err;
+ if (c->opts.verbose || !c->sb.clean)
+ bch_info(c, "journal replay done");
- ret = bch2_check_lrus(c, true);
+ bch_info(c, "checking need_discard and freespace btrees");
+ ret = bch2_check_alloc_info(c);
if (ret)
goto err;
bch_verbose(c, "done checking need_discard and freespace btrees");
- }
- bch2_stripes_heap_start(c);
+ set_bit(BCH_FS_CHECK_ALLOC_DONE, &c->flags);
- clear_bit(BCH_FS_REBUILD_REPLICAS, &c->flags);
- set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
- set_bit(BCH_FS_MAY_GO_RW, &c->flags);
-
- /*
- * Skip past versions that might have possibly been used (as nonces),
- * but hadn't had their pointers written:
- */
- if (c->sb.encryption_type && !c->sb.clean)
- atomic64_add(1 << 16, &c->key_version);
-
- if (c->opts.norecovery)
- goto out;
+ bch_info(c, "checking lrus");
+ ret = bch2_check_lrus(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "done checking lrus");
+ set_bit(BCH_FS_CHECK_LRUS_DONE, &c->flags);
- bch_verbose(c, "starting journal replay, %zu keys", c->journal_keys.nr);
- err = "journal replay failed";
- ret = bch2_journal_replay(c);
- if (ret)
- goto err;
- if (c->opts.verbose || !c->sb.clean)
- bch_info(c, "journal replay done");
+ bch_info(c, "checking backpointers to alloc keys");
+ ret = bch2_check_btree_backpointers(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "done checking backpointers to alloc keys");
- err = "error initializing freespace";
- ret = bch2_fs_freespace_init(c);
- if (ret)
- goto err;
+ bch_info(c, "checking backpointers to extents");
+ ret = bch2_check_backpointers_to_extents(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "done checking backpointers to extents");
- if (c->opts.fsck) {
- bch_info(c, "checking alloc to lru refs");
- err = "error checking alloc to lru refs";
- ret = bch2_check_alloc_to_lru_refs(c);
+ bch_info(c, "checking extents to backpointers");
+ ret = bch2_check_extents_to_backpointers(c);
if (ret)
goto err;
+ bch_verbose(c, "done checking extents to backpointers");
+ set_bit(BCH_FS_CHECK_BACKPOINTERS_DONE, &c->flags);
- ret = bch2_check_lrus(c, true);
+ bch_info(c, "checking alloc to lru refs");
+ ret = bch2_check_alloc_to_lru_refs(c);
if (ret)
goto err;
bch_verbose(c, "done checking alloc to lru refs");
- }
+ set_bit(BCH_FS_CHECK_ALLOC_TO_LRU_REFS_DONE, &c->flags);
+ } else {
+ set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_ALLOC_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_LRUS_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_BACKPOINTERS_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_ALLOC_TO_LRU_REFS_DONE, &c->flags);
+ set_bit(BCH_FS_FSCK_DONE, &c->flags);
- if (c->sb.version < bcachefs_metadata_version_snapshot_2) {
- bch2_fs_lazy_rw(c);
+ if (c->opts.norecovery)
+ goto out;
- err = "error creating root snapshot node";
- ret = bch2_fs_initialize_subvolumes(c);
+ set_bit(BCH_FS_MAY_GO_RW, &c->flags);
+
+ bch_verbose(c, "starting journal replay, %zu keys", c->journal_keys.nr);
+ ret = bch2_journal_replay(c, last_seq, blacklist_seq - 1);
if (ret)
goto err;
+ if (c->opts.verbose || !c->sb.clean)
+ bch_info(c, "journal replay done");
}
- bch_verbose(c, "reading snapshots table");
- err = "error reading snapshots table";
- ret = bch2_fs_snapshots_start(c);
+ ret = bch2_fs_freespace_init(c);
if (ret)
goto err;
- bch_verbose(c, "reading snapshots done");
+
+ if (c->sb.version < bcachefs_metadata_version_bucket_gens &&
+ c->opts.version_upgrade) {
+ bch_info(c, "initializing bucket_gens");
+ ret = bch2_bucket_gens_init(c);
+ if (ret)
+ goto err;
+ bch_verbose(c, "bucket_gens init done");
+ }
if (c->sb.version < bcachefs_metadata_version_snapshot_2) {
- /* set bi_subvol on root inode */
- err = "error upgrade root inode for subvolumes";
- ret = bch2_trans_do(c, NULL, NULL, BTREE_INSERT_LAZY_RW,
- bch2_fs_upgrade_for_subvolumes(&trans));
+ ret = bch2_fs_upgrade_for_subvolumes(c);
if (ret)
goto err;
}
if (c->opts.fsck) {
- bch_info(c, "starting fsck");
- err = "error in fsck";
ret = bch2_fsck_full(c);
if (ret)
goto err;
bch_verbose(c, "fsck done");
} else if (!c->sb.clean) {
bch_verbose(c, "checking for deleted inodes");
- err = "error in recovery";
ret = bch2_fsck_walk_inodes_only(c);
if (ret)
goto err;
le16_to_cpu(c->sb.version_min) < bcachefs_metadata_version_btree_ptr_sectors_written) {
struct bch_move_stats stats;
- bch_move_stats_init(&stats, "recovery");
+ bch2_move_stats_init(&stats, "recovery");
bch_info(c, "scanning for old btree nodes");
- ret = bch2_fs_read_write(c);
- if (ret)
- goto err;
-
- ret = bch2_scan_old_btree_nodes(c, &stats);
+ ret = bch2_fs_read_write(c) ?:
+ bch2_scan_old_btree_nodes(c, &stats);
if (ret)
goto err;
bch_info(c, "scanning for old btree nodes done");
set_bit(BCH_FS_FSCK_DONE, &c->flags);
bch2_flush_fsck_errs(c);
- if (!c->opts.keep_journal) {
+ if (!c->opts.keep_journal &&
+ test_bit(JOURNAL_REPLAY_DONE, &c->journal.flags)) {
bch2_journal_keys_free(&c->journal_keys);
bch2_journal_entries_free(c);
}
kfree(clean);
+
+ if (!ret && test_bit(BCH_FS_HAVE_DELETED_SNAPSHOTS, &c->flags)) {
+ bch2_fs_read_write_early(c);
+ bch2_delete_dead_snapshots_async(c);
+ }
+
if (ret)
- bch_err(c, "Error in recovery: %s (%i)", err, ret);
+ bch_err_fn(c, ret);
else
- bch_verbose(c, "ret %i", ret);
+ bch_verbose(c, "ret %s", bch2_err_str(ret));
return ret;
err:
fsck_err:
struct bch_inode_unpacked root_inode, lostfound_inode;
struct bkey_inode_buf packed_inode;
struct qstr lostfound = QSTR("lost+found");
- const char *err = "cannot allocate memory";
struct bch_dev *ca;
unsigned i;
int ret;
c->disk_sb.sb->compat[0] |= cpu_to_le64(1ULL << BCH_COMPAT_extents_above_btree_updates_done);
c->disk_sb.sb->compat[0] |= cpu_to_le64(1ULL << BCH_COMPAT_bformat_overflow_done);
+ if (c->sb.version < bcachefs_metadata_version_inode_v3)
+ c->opts.version_upgrade = true;
+
if (c->opts.version_upgrade) {
c->disk_sb.sb->version = cpu_to_le16(bcachefs_metadata_version_current);
c->disk_sb.sb->features[0] |= cpu_to_le64(BCH_SB_FEATURES_ALL);
mutex_unlock(&c->sb_lock);
set_bit(BCH_FS_INITIAL_GC_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_LRUS_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_BACKPOINTERS_DONE, &c->flags);
+ set_bit(BCH_FS_CHECK_ALLOC_TO_LRU_REFS_DONE, &c->flags);
set_bit(BCH_FS_MAY_GO_RW, &c->flags);
set_bit(BCH_FS_FSCK_DONE, &c->flags);
for_each_online_member(ca, c, i)
bch2_dev_usage_init(ca);
- err = "unable to allocate journal buckets";
for_each_online_member(ca, c, i) {
ret = bch2_dev_journal_alloc(ca);
if (ret) {
bch2_fs_journal_start(&c->journal, 1);
bch2_journal_set_replay_done(&c->journal);
- err = "error going read-write";
ret = bch2_fs_read_write_early(c);
if (ret)
goto err;
* btree updates
*/
bch_verbose(c, "marking superblocks");
- err = "error marking superblock and journal";
for_each_member_device(ca, c, i) {
ret = bch2_trans_mark_dev_sb(c, ca);
if (ret) {
ca->new_fs_bucket_idx = 0;
}
- bch_verbose(c, "initializing freespace");
- err = "error initializing freespace";
ret = bch2_fs_freespace_init(c);
if (ret)
goto err;
- err = "error creating root snapshot node";
ret = bch2_fs_initialize_subvolumes(c);
if (ret)
goto err;
bch_verbose(c, "reading snapshots table");
- err = "error reading snapshots table";
ret = bch2_fs_snapshots_start(c);
if (ret)
goto err;
bch_verbose(c, "reading snapshots done");
- bch2_inode_init(c, &root_inode, 0, 0,
- S_IFDIR|S_IRWXU|S_IRUGO|S_IXUGO, 0, NULL);
+ bch2_inode_init(c, &root_inode, 0, 0, S_IFDIR|0755, 0, NULL);
root_inode.bi_inum = BCACHEFS_ROOT_INO;
root_inode.bi_subvol = BCACHEFS_ROOT_SUBVOL;
- bch2_inode_pack(c, &packed_inode, &root_inode);
+ bch2_inode_pack(&packed_inode, &root_inode);
packed_inode.inode.k.p.snapshot = U32_MAX;
- err = "error creating root directory";
ret = bch2_btree_insert(c, BTREE_ID_inodes,
&packed_inode.inode.k_i,
NULL, NULL, 0);
- if (ret)
+ if (ret) {
+ bch_err_msg(c, ret, "creating root directory");
goto err;
+ }
bch2_inode_init_early(c, &lostfound_inode);
- err = "error creating lost+found";
ret = bch2_trans_do(c, NULL, NULL, 0,
bch2_create_trans(&trans,
BCACHEFS_ROOT_SUBVOL_INUM,
0, 0, S_IFDIR|0700, 0,
NULL, NULL, (subvol_inum) { 0 }, 0));
if (ret) {
- bch_err(c, "error creating lost+found");
+ bch_err_msg(c, ret, "creating lost+found");
goto err;
}
goto err;
}
- err = "error writing first journal entry";
ret = bch2_journal_flush(&c->journal);
- if (ret)
+ if (ret) {
+ bch_err_msg(c, ret, "writing first journal entry");
goto err;
+ }
mutex_lock(&c->sb_lock);
SET_BCH_SB_INITIALIZED(c->disk_sb.sb, true);
return 0;
err:
- pr_err("Error initializing new filesystem: %s (%i)", err, ret);
+ bch_err_fn(ca, ret);
return ret;
}
projects / bcachefs-tools-debian / blobdiff